Patrick Vlaskovits (Go Falcons!) was over at MuckerLab the other day talking to our companies about customer development which reminded me of the most important (only?) skill I took away in business school – segmentation. While “strategy” is the most over used word in business, “segmentation” is probably the most underutilized. It’s certainly not as sexy as “strategy” but it is the foundation from which any business is built. In a world of limited resources – a laser like focus on a few worthy target customers will help create efficiencies and purpose across the entire company from marketing, advertising, product management, engineering, – and yes, strategy.
The below are just some thoughts from Patrick’s presentation and my own experiences.
Choosing Segmentation Attributes
Its very easy to start the segmentation exercise thinking in terms of demographic information (my target customer is an Asian American male, 24-35, earning $50K a year etc ). Segmenting using these attributes is at best, lazy; and at worst, useless.
- Does the difference in these attributes (male vs. female) explain the varying needs or pain points of each of the segments? E.g. I want widget X cause its cheaper vs. I want widget X cause it makes me look cool? If not, don’t use them.
- Can my customers self identify through these attributes? Self discoverable attributes allows you (the startup) to qualify a customer/user through a simple questionnaire at the beginning of the engagement/funnel before investing too much resources in attempting to acquire that customer (or user).
- Are the attributes unique to the market you are targeting? If your market is the offset printing industry – talking about demographic information of the owner might not be an useful exercise.
It is really hard to get started in any segmentation exercise – often the best way to build some momentum is to understand who are NOT your customers. In most cases, at the end of the exercise you should have significantly more “anti-segments” than “target segments” – otherwise you haven’t really made the hard choice of selecting your customers and truly understanding who your best users/customers might be.
The McKinsey “MECE” Test
Make sure your segmentation scheme (Anti Segments + Target Segments) is “Mutually Exclusive and Collectively Exhaustive.” Its really bad to “miss” a segment because 1) that might be the segment you should be targeting 2) if you or someone at your company come across an user in that segment but put them in the wrong bucket (cause nothing fits) it will actually give you false readings for your hypothesis driven market tests.
Iteration & Segmentation
I like teams that are able to iterate at a high velocity. However; all too often, I see companies endlessly pivoting from market to market without really understanding WHY their initial product didn’t achieve market fit in the first place. One of the area to dig into as part of the iteration exercise is to “re-do” your segmentation scheme. If the user feedback on why they did not adopt the product is inconsistent WITHIN a segment, its highly likely you’ve screwed up the segmentation in the first place. As a result, instead of revamping product drastically, or pivoting to a whole new market – try re-segmenting the target customer base, rebuild your value hypothesis for a new segment, and make it another go.
Acquisition Channel & Segmentation – The “Usefulness” Test
Often, the created segmentation scheme looks good on paper but ends up pretty useless in practice. When there are no efficient, obvious, and cheap acquisition channels to reach a particular segment – its better to start over than to over compensate with impractical or expensive acquisition campaigns. Large companies have the marketing budget to only reach 25% or less of their intended segment with a mass market campaigns – startups don’t have that luxury and need to hit at least 75% efficiency on its acquisition spend. In terms of velocity and iteration testing – it will be next to impossible to get the right sample size, speed, and feedback loop if the acquisition channels do not match nicely with your segmentation scheme. This is why I typically stay away from using psychographic profiles as part of a segmentation model – there is just no simple or efficient marketing channels to reach, as an example, “upwardly mobile jet setting urban gen X’ers.”